I'd rather not think too hard about 3.3.x any time soon :) but, yes, we'd have to take care with the workflow. At this time, I don't think we should worry too heavily about maintaining more than two lines of releases at a time which is what we've been doing thus far with 3.1.x (tp31) and 3.2.x (master). I think we should continue with that pattern for a while where after this release we do 3.1.3 on tp31 and 3.2.1 on master taking care to focus on non-breaking change for both release branches. Then the merge flow stays the same as what we've been doing. If there are more opinions about this, please start a fresh DISCUSS thread and reference this thread so we can keep this current thread more focused on code freeze/release issues.
On Tue, Apr 5, 2016 at 11:30 AM, Dylan Millikin <dylan.milli...@gmail.com> wrote: > I agree. We have been merging upstream so it's only natural that 3.1.2 be > released before 3.2.0. I was a little confused about having 3.2.0 come out > before so now it makes more sense. > > If in the future we want to be able to do this we will need our workflow to > merge downstream instead. Basically that would mean that all changes we > want to make to 3.2.1 would be done against 3.3.0 then cherry picked and > merged down to 3.2.1 (There's a subtle difference, if you fix a feature on > 3.2.1 that no longer exists in 3.3.0 for example). > As far as the changelogs go, you would add the changes to whichever > version(s) they were merged/applied to even if it means having duplicates. > The changes merged from 3.3.0 to 3.2.1 would be in the changelog for both > version but 3.2.1 would have an extra mention like "backport" (which would > most likely be the majority of changes). > Honestly this is tedious and only worth it if we plan on providing extended > support for minor versions, (do we want to? guess this would warrant it's > own discussion anyways). > > > On Tue, Apr 5, 2016 at 3:58 PM, Marko Rodriguez <okramma...@gmail.com> > wrote: > > > Hi, > > > > I think we should release 3.1.2 as well. > > > > Marko. > > > > http://markorodriguez.com > > > > On Apr 5, 2016, at 7:56 AM, Stephen Mallette <spmalle...@gmail.com> > wrote: > > > > > I was reviewing the upgrade docs and release notes today and realized > we > > > have some weirdness because TinkerPop 3.2.0 is releasing prior to > 3.1.2. > > > 3.2.0 encompasses all of the changes in 3.1.2, so to find out what > 3.2.0 > > > has, you kinda have to look at both 3.1.2 and 3.2.0 upgrade docs. I'm > > > starting to wonder if that will be confusing for folks who scroll down > to > > > 3.1.2 to see "Not Officially Released Yet". > > > > > > Marko had asked at one point if we were releasing 3.1.2 along with > 3.2.0 > > > and I'd indicated an answer of "no", but looking at it this way makes > me > > > wonder if that's the right call. It seems like the call to release a > > > downstream version of TinkerPop should trigger the release of all > > versions > > > that it encompasses. So I guess the question is whether or not we > should: > > > > > > 1. release 3.1.2 in conjunction with 3.2.0 (3.1.2 is as ready to go imo > > as > > > 3.2.0 at this point) > > > 2. make it a TinkerPop policy to release all dependent versions of the > > most > > > recent expected release > > > > > > Of course, this does mean that we need to focus on testing BOTH 3.1.2 > and > > > 3.2.0 this week if we want to go this route so there's some added work > > > there. Thoughts? > > > > > > On Mon, Apr 4, 2016 at 12:34 PM, Dylan Millikin < > > dylan.milli...@gmail.com> > > > wrote: > > > > > >> Sounds good. > > >> > > >> On Mon, Apr 4, 2016 at 6:33 PM, Stephen Mallette < > spmalle...@gmail.com> > > >> wrote: > > >> > > >>> I think we should basically freeze the whole repo at this point. I'd > > said > > >>> in the last post that tp31 branch was still open to dev, but that's > > not a > > >>> great idea as we might yet have tweaks for master that could occur in > > >>> tp31. So, I think the better approach should be to assume that > master > > >> and > > >>> tp31 are both frozen except for change that will go into 3.2.0's > > release > > >>> build this friday. > > >>> > > >>> On Fri, Apr 1, 2016 at 2:00 PM, Stephen Mallette < > spmalle...@gmail.com > > > > > >>> wrote: > > >>> > > >>>> Code freeze is basically in effect starting tomorrow for our master > > >>>> branch. Development on the tp31 branch can continue as needed, but, > > of > > >>>> course, do not merge tp31 back to master during our freeze. Please > use > > >>> this > > >>>> week to run tests and report problems/findings. > > >>>> > > >>>> As usual it would be great to hear from driver/graph providers next > > >> week > > >>>> to see how their implementations are working against 3.2.0-SNAPSHOT > (I > > >>> will > > >>>> publish a "final" SNAPSHOT later today for testing). > > >>>> > > >>>> I would have liked to have gotten this PR from Kuppitz merged before > > >>>> freeze: > > >>>> > > >>>> https://github.com/apache/incubator-tinkerpop/pull/286 > > >>>> > > >>>> as that's a really good change, but it really isn't required for our > > >>>> "release" - it's for our development productivity, so i don't think > we > > >>> need > > >>>> to rush for that. > > >>>> > > >>>> Thanks, > > >>>> > > >>>> Stephen > > >>>> > > >>> > > >> > > > > >
